R Notebook

reading the train file and see it column name

source("CreditDefaultUtils.R")
library(tidyverse)
previousApplicationData <- read_csv("data/previous_application.csv")
installmentsPpaymentsData <- read_csv("data/installments_payments.csv")
creditCardBalanceData <- read_csv("data/credit_card_balance.csv")
POSCASHBalanceData <- read_csv("data/POS_CASH_balance.csv")

library(dplyr)
previousApplicationData %>% summarise_all(funs(n_distinct(.)))

PACat <- getCatFeatures(previousApplicationData)
PACat

PACatNoNa <- PACat[, -which(colMeans(is.na(PACat)) > 0.9)]
PACatNoNa

PANum <- getNumFeatures(previousApplicationData,'SK_ID_PREV')
PANum

previousApplicationReady <- PANum %>% inner_join(PACatNoNa)
previousApplicationReady

write.csv(previousApplicationReady, file = "data/previous_application_ready.csv",row.names=FALSE)

library(dplyr)
installmentsPpaymentsData %>% summarise_all(funs(n_distinct(.)))

IPNum <- getNumFeatures(installmentsPpaymentsData,'SK_ID_PREV')
IPNum

write.csv(IPNum, file = "data/installments_payments_ready.csv",row.names=FALSE)

library(dplyr)
creditCardBalanceData %>% summarise_all(funs(n_distinct(.)))

CBCat <- getCatFeatures(creditCardBalanceData)
CBCatNoNa <- CBCat[, -which(colMeans(is.na(CBCat)) > 0.9)]
CBCatNoNa

CBNum <- getNumFeatures(creditCardBalanceData,'SK_ID_PREV')
CBNum

creditCardBalanceDataReady <- CBNum %>% inner_join(CBCatNoNa)
creditCardBalanceDataReady

write.csv(creditCardBalanceDataReady, file = "data/credit_card_balance_ready.csv",row.names=FALSE)

PCCat <- getCatFeatures(POSCASHBalanceData)
PCCatNoNa <- PCCat[, -which(colMeans(is.na(PCCat)) > 0.9)]
PCCatNoNa

PCNum <- getNumFeatures(POSCASHBalanceData,'SK_ID_PREV')
PCNum

POSCASHBalanceDataReady <- PCNum %>% inner_join(PCCatNoNa)
POSCASHBalanceDataReady

write.csv(POSCASHBalanceDataReady, file = "data/POS_CASH_Balance_Ready.csv",row.names=FALSE)